Genesys Cloud - Developer Community!

 View Only

Sign Up

  • 1.  Real Time Voice Translation

    Posted 5 days ago

    Hi Team,

     

    I need some help figuring out the right approach for real-time voice translation during active agent conversations.

     

    What I'm Trying to Achieve:

     

    When a Spanish-speaking (Any language) customer is talking to an English-speaking agent:

    • Customer speaks Spanish → Agent hears English (translated in real-time)
    • Agent speaks English → Customer hears Spanish (translated in real-time)

    I've got all the translation tech working (speech-to-text, translation, text-to-speech), but I'm struggling with how to integrate it into Genesys so it works during a live agent conversation.

     

    What I've Tried:

     

    Attempt 1: Audio Connector

    • Connects successfully with bidirectional audio
    • Translation works perfectly
    • Problem: The Architect flow stops at the Audio Connector block and never reaches the agent
    • When I send a disconnect message to continue the flow, the AudioHook session closes
    • So translation ends before the agent even answers

    Attempt 2: AudioHook Monitor

    • Flow works normally, agent can answer
    • I can transcribe what the customer says and show translation text to the agent
    • Problem: It's receive-only - I can't play translated audio back to the customer
    • So the customer still hears the agent speaking English (not helpful!)

    The Core Problem:

     

    I need bidirectional audio (so both parties hear translations), but I also need the call to route normally to an agent.

     

    Audio Connector gives me bidirectional audio but blocks the flow. AudioHook Monitor lets the flow continue but can't play audio back.

     

    My Questions:

    1. Is there a way to use Audio Connector without it blocking the Architect flow?
    2. Is there a Genesys API to play audio into an active agent conversation (while using AudioHook Monitor for listening)?
    3. Am I thinking about this wrong - is there a different Genesys feature I should be using for live agent translation?

    #Integrations

    ------------------------------
    Paul Smith
    Senior Engineer
    ------------------------------


  • 2.  RE: Real Time Voice Translation

    Posted 3 days ago

    Hello,

    As far as I know, it is not possible to achieve this With AudioConnector nor AudioHook Monitor (as you already found out).

    I see 2 Genesys Cloud ideas sort of related to such scenario:

    The only approach I can think of at this time (with current features) would be to bridge two SIP calls.

    A call that reaches Genesys Cloud and is transferred (with bridging) to your infrastructure. And another call from your infrastructure to Genesys Cloud. Which could allow you to manage audio streams in and out. The drawback that I can think of is possibly with ANI/CLI (unless you have a direct SIP connection to Genesys Cloud from your infra - to preserve customer phone number in the second call).

    Regards,



    ------------------------------
    Jerome Saint-Marc
    Senior Development Support Engineer
    ------------------------------